Yunan Luo
Georgia Institute of Technology, GA·Computer Science·Early-career
Proteins carry out most of the work inside living cells, yet scientists know what only a small share of them do. Research has long focused on a few well-studied proteins, which leaves most proteins poorly understood.
- Fields
Computer Science
Biochemistry
Biomedical Engineering
- Topics
protein function prediction
protein function annotation
annotation bias
- Grants
- 3 known grants, through Aug 2030
What they work on
- Develop machine learning methods that learn protein representations from amino acid sequence and 3D structure, in order to explain how composition and shape determine protein function.
- Build biology-guided deep learning frameworks that include evolutionary and structural knowledge to predict protein function across whole genomes and to predict how mutations change protein function.
- Create a platform combining data-efficient models, uncertainty-guided exploration and generative models for AI-aided directed evolution and joint design of protein sequence and structure.
- Design ML models for protein function annotation that correct for biased and sparse training data, so that understudied proteins get accurate predictions with controlled error rates.
- Develop evolution-guided generative AI and protein language models to design proteins that balance multiple properties when little experimental data is available.
- Couple AI protein design with an automated biofoundry that builds and tests the designed proteins in a closed loop, using the results to refine later designs.
More about the research
Designing better proteins for medicine and industry is also slow and costly. The three projects described here build artificial intelligence (AI) and machine learning (ML) methods that link a protein's sequence and three-dimensional (3D) shape to its function. The methods then use these links to predict what unknown proteins do and to design new ones.
Why it matters
Better predictions of protein function and faster protein design could speed up drug discovery, vaccine development and personalized diagnosis and treatment. Automated, AI-guided protein engineering could also lower the cost of making useful proteins for biotechnology, energy, agriculture and chemical manufacturing. Openly released software would let other labs apply these tools to understudied proteins.
Known grants
From NIH and NSF research grants. Labs often have other funding we can't see.
CAREER: Machine Learning Algorithms for Tackling Annotation Inequality in Protein Function Characterization
NSF · Sep 2025 – Aug 2030
Integrative deep learning algorithms for understanding protein sequence-structure-function relationships: representation, prediction, and discovery
NIH · Sep 2023 – Jul 2028
Collaborative Research: ACED: Accelerating Protein Engineering with Evolution-Guided Generative AI and a Self-Driving Biofoundry
NSF · Jul 2025 – Dec 2026
Closest labs
Rama Ranganathan
University of Chicago, IL·Biophysics
Develop and experimentally validate a deep generative model that enables text-prompted engineering of diverse, functional artificial proteins.
Junzhou Huang
University of Texas at Arlington, TX·Computer Science
Develop multi-modal deep learning methods, including large language models, to predict the functions of thousands of poorly annotated genes and support analysis of gene-gene synergies.
Jesse Thaler
Massachusetts Institute of Technology, MA·Physics
Develop "ab initio AI" methods that build first principles of fundamental physics into machine learning, to make intractable theoretical physics calculations, such as precision structure-of-matter predictions, feasible.
The school
Georgia Institute of Technology: 50 labs on PI Pages, $1.5B a year on research, 572 research doctorates a year.
Reach out
A start for your email. Fill in the brackets.